differential-review

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted code diffs and pull request data, which can contain instructions or data designed to subvert the analysis process. This is an inherent risk for auditing tools.
  • Ingestion points: Untrusted source code and metadata are read using git diff, gh pr view, and Read tools during the analysis phases.
  • Boundary markers: The skill relies on structured analysis phases (Phases 0-6) but lacks explicit delimiters or instructions to ignore embedded commands in the reviewed code.
  • Capability inventory: The agent has access to Bash, Read, and Write tools, enabling it to execute commands and generate report files.
  • Sanitization: Code content is processed as-is to ensure an accurate security review.
  • [COMMAND_EXECUTION]: The skill methodology requires the use of shell commands to extract codebase history and statistics.
  • Evidence: The use of git log, git diff, git blame, find, wc, and grep is standard for the described security review workflow.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 06:43 AM
Security Audit — agent-trust-hub — differential-review